摘要
针对视频LSB匹配隐写,提出了一种基于区域相关性原理,通过构造区域相关性图直接提取载体或载密视频帧特征,应用中值滤波去除不同视频载体之间的特征差异,逐帧进行视频隐写分析的方法。实验结果表明,该算法能有效检测每帧均进行嵌入且嵌入率仅为p=0.1时的视频,而且在只对20%的视频帧进行混合嵌入的情况下,也能达到90.91%的正确检测率。
In this paper,a regional correlation-based principle was proposed according to the video LSB matching steganographyaocal.By constructing the images of regional correlation,the features were extracted from carriers or stego-video frequency frames,and then,median filter was applied to remove the differences of features between different carriers.Experimental results showed that the algorithm could detect not only all video frequency frames were embedded with the embedding rate of p=0.1 effectively,but also the accuracy could be 90.91% under the situation of only 20% of video frequency frames were made with mixed embedding.
出处
《山东科技大学学报(自然科学版)》
CAS
2011年第1期62-67,共6页
Journal of Shandong University of Science and Technology(Natural Science)
基金
国家重点基础研究发展计划("973"计划)前期研究专项(2010CB334706)
国家自然科学基金项目(60702065
61072122)